39Effectuation has been studied relatively well in the context of Knightian uncertainty, a term originating from Frank Knight’s taxonomy of uncertainty in his 1921 thesis, Risk, Uncertainty, and Profit. In lay terms, Knightian uncertainty refers to situations in which the future is not only unknown but also fundamentally unknowable. An iconic example from decision theory can help clarify Knight’s taxonomy. Imagine you are playing a game in which you draw balls from an urn containing 50 green balls and 50 red balls. You will win if you draw a green ball. Although you do not know which ball you will draw, you can still calculate the odds as 50–50 since you know the distribution of balls in the urn. This captures the idea of “risk”—namely, a known set of possibilities but an unknown draw.

Another concept of interest is the notion of “uncertainty” in which you know neither the distribution nor the draw. This would be like an urn containing many different colored balls, but you do not know how many of each color or even the total. The game, however, is the same: You win if you draw a green ball. It is easy to see that this game is much more difficult to play than the game of risk. Many organizational, economic, and socio-political problems are conceptualized as S. D. Sarasvathy

40problems of uncertainty that can only be tackled through sophisticated techniques for prediction ranging from systematic hypothesis-testing to scenario analysis and other approaches based on simulation and big data.

In both the above thought experiments, we knew something about the urn’s contents. In situations in which Knightian uncertainty is involved, even this information is unavailable. The urn may contain things that defy classification or even recognition, making it impossible to classify them into a distribution on which predictive techniques can work. It is as though the urn could contain umbrellas, snakes, bars of gold, disease, anything and everything that can and may exist. You get something different every time you draw—not just balls. In other words, Knightian uncertainty refers to the impossibility of imagining, let alone specifying a distribution, on the basis of which you can make predictions. In dealing with Knightian uncertainty, you need to come up with techniques that either minimize or completely avoid prediction altogether. The lessons that expert entrepreneurs learn consist in nonpredictive techniques that we call effectuation or effectual logic, contrasted with predictive or causal logic.

Effectuators develop an awareness of and even a preference for Knightian uncertainty. Hence, in addition to cocreating futures with self-selected stakeholders, effectual approaches emphasize possible errors as decision criteria rather than predicted upsides (e.g., the affordable loss principle). This is a powerful tool to help bring downsides within one’s control, without constraining upsides. Therefore,